Bill Gates warns of two decades of turbulent AI disruption

Bill Gates, the cofounder and former CEO of Microsoft and chairman of the Gates Foundation, stated that society must prepare for approximately 20 years of turbulent disruption driven by artificial intelligence. In a recent interview, Gates described the coming era as potentially splitting society between those comfortable living near the technological frontier and those who reject it.

Gates noted that the current policy response from the White House and the lack of coordinated efforts from the AI industry are insufficient to manage the risks posed by the technology. He argued that the progress over the last year, including the ability of models to write code, find cyber flaws, reason in biology, and form relationships, means that every one of these capabilities will materialize and society needs to be ready. Gates emphasized that while the good effects of AI may arrive slowly, the negative consequences will not be minimized if the technology is not managed correctly.

The former Microsoft CEO identified models without safeguards as the primary near-term risk, distinguishing them from models where providers actively monitor usage to prevent exploitation by cyberattackers or bioterrorists. Gates stated that the basic power of these models toward social relationships, the job market, and security threats constitutes the most dangerous technology ever invented. He observed that while industry insiders universally acknowledge these dangers, awareness outside the sector remains low, with many dismissing current warnings as mere hype or ineffective public relations.

Gates rejected the notion that he is an AI doomer, instead positioning himself as a strong voice for the profound nature of the technology. He highlighted that AI will disrupt every area of human activity, moving from white-collar intellectual work to physical blue-collar tasks through humanoid robots. The former Microsoft CEO criticized the current regulatory environment, noting that with multiple competing firms at the frontier, responsibility for safety standards must shift to the government, similar to how other dangerous products are regulated. He expressed concern about the government's ability to understand the technology it seeks to regulate but maintained that mandatory review of models and the creation of safeguards are necessary steps.

In response to accelerationist views, Gates reacted sharply to comments by Ben Horowitz, who suggested that safety research at Anthropic was manifesting human extinction. Gates called the claim bizarre and denied any involvement in gain-of-function research, a topic Horowitz had linked to pandemic preparation efforts. Gates agreed with Mustafa Suleyman that focusing on the moral standing or consciousness of AI should not be a priority. He argued that the immediate challenges for humans, including job market disruption and infrastructure attacks, require focusing on human welfare first. Gates concluded that while control problems are significant, they are not imminent compared to the immediate threats posed by bad actors using current models.